Oxidative Regulation of Large Conductance Calcium-Activated Potassium Channels

Large conductance Ca²⁺-activated K⁺ channels (BK_Ca or Slo) are ubiquitously present and play a variety of physiological roles, generally providing an inhibitory negative feedback influence that links cellular metabolism and excitability (Marty 1989; McManus 1991; Toro and Stefani 1991; Jan and Jan 1997; Toro et al. 1998; Vergara et al. 1998). The physiological importance of BK_Ca channels has been well documented in regulation of the vascular tone (Brayden 1996; Rusch et al. 1996; Brenner et al. 2000; Pluger et al. 2000), neurosecretion (Lingle et al. 1996), and cochlear frequency tuning (Fettiplace and Fuchs 1999; Ramanathan et al. 1999).

BK_Ca or Slo channels are characterized by their large single-channel openings (>200 pS in symmetrical 100 mM KCl) and voltage-dependent activation modulated by intracellular Ca²⁺ (Latorre et al. 1984; Pallotta et al. 1992; Kaczorowski et al. 1996; Meera et al. 1996; Cui et al. 1997; Jan and Jan 1997; Atkinson et al. 1998; Vergara et al. 1998; Horrigan et al. 1999). Although its activation is favored by Ca²⁺, the Slo channel may be fully activated by depolarization alone without Ca²⁺ (Cui et al. 1997). The amino acid sequences of the Slo α subunits, originally isolated from Drosophila (Atkinson et al. 1991; Adelman et al. 1992), indicate that each subunit has a “core” domain and the COOH “tail” domain. The structural organization of the core domain is similar to that of Shaker-type voltage-gated Kv channels, in that it contains six putative transmembrane segments (S1–S6) and the pore (P) segment. The S4 segment contains several positively charged amino acid residues that are involved in voltage-dependent gating of the Slo channel (Noceti et al. 1996; Stefani et al. 1997; Diaz et al. 1998). The Slo core domain also includes the S0 segment in the NH₂ terminus, such that the NH₂-terminal end faces the extracellular side (Meera et al. 1997). This domain at least, in part, mediates the interaction of Slo with the auxiliary β subunit (Wallner et al. 1996), which increases the overall activity of the channel largely in a Ca²⁺-independent manner (Nimigean and Magleby 2000). The long tail domain of the Slo subunit is considered to be cytoplasmic (Meera et al. 1997) and contributes to the Ca²⁺ sensitivity of the channel (Wei et al. 1994; Schreiber and Salkoff 1997), possibly by forming a Ca²⁺ binding motif termed the calcium bowl (Schreiber and Salkoff 1997). Within the tail domain, multiple subdomains may be involved in mediating the Ca²⁺ sensitivity (Schreiber et al. 1999). As found with voltage-gated Kv channels, it is likely that four Slo subunits form one functional channel (Shen et al. 1994) with or without β subunits depending on the cell type (Chang et al. 1997; Tanaka et al. 1997; Wanner et al. 1999).

Gating of BK_Ca channels has been extensively studied both in native and heterologous expression systems. Single-channel studies show that the Slo or BK_Ca channel displays bursts of openings with short flicker closures involving multiple dwell time components (Barrett et al. 1982; Methfessel and Boheim 1982; Moczydlowski and Latorre 1983; Singer and Walsh 1987; McManus and Magleby 1988, McManus and Magleby 1991; Song and Magleby 1994; Rothberg and Magleby 1998, Rothberg and Magleby 1999). Heterologous expression of Slo α subunits has allowed the combined use of single-channel, macroscopic ionic current and macroscopic gating current measurements to better understand the Slo channel gating behavior (Cox et al. 1997b; Cui et al. 1997; Stefani et al. 1997; Diaz et al. 1998; Horrigan and Aldrich 1999; Horrigan et al. 1999). Many aspects of voltage- and Ca²⁺-dependent gating properties of Slo channels have been successfully modeled by allosteric voltage-dependent schemes incorporating probable structural features of the Slo channel, such as the formation of homomultimeric tetramers (Cox et al. 1997a; Horrigan and Aldrich 1999; Horrigan et al. 1999; Rothberg and Magleby 1999). However, a full and complete description of the channel gating behavior probably requires a model with additional transitions (Nimigean and Magleby 1999; Talukder and Aldrich 2000).

Reactive oxygen/nitrogen species (ROS/RNS) are commonly generated in vivo and implicated in many physiological functions and pathological conditions such as Alzheimer's disease and Parkinson's disease by oxidatively modifying various proteins (Halliwell 1992; Evans 1993; Jenner and Olanow 1996; Markesbery 1997; Stadtman and Berlett 1998). ROS/RNS, capable of altering various ion transport mechanisms (Kourie 1998), are considered to play critical roles in the regulation of vascular tension (Rubanyi 1988) and also in reperfusion injury (Opie 1989). Because of their importance in controlling vascular tone (Brayden 1996; Rusch et al. 1996; Brenner et al. 2000; Pluger et al. 2000), the regulation of BK_Ca channels by nitric oxide (NO) has been extensively studied (Beech 1997; Michelakis et al. 1997). Effects of NO mediated by cGMP-dependent signal transduction pathways are well-known (Southam and Garthwaite 1996; Vaandrager and de Jonge 1996; Lohmann et al. 1997). NO (or its related species) may also regulate its physiological effectors by acting as a weak radical, promoting oxidation (Stamler 1994). In rat pituitary, NO enhances the BK_Ca activity in a guanylyl cyclase–independent manner, suggesting that RNS may directly modify the BK_Ca protein by acting on sulfhydryl groups (Ahern et al. 1999). Oxidizing agents such as H₂O₂ may induce up- or downregulation of BK_Ca channels, depending on the experimental preparation used (Park et al. 1995; Thuringer and Findlay 1997; Wang et al. 1997; Barlow and White 1998; Hayabuchi et al. 1998; Brzezinska et al. 2000; Gong et al. 2000). DiChirara and Reinhart (1997) showed that H₂O₂ decreased the activity of the human Slo (hSlo) channel by shifting its voltage sensitivity to a more positive direction (Dichiara and Reinhart 1997). They also showed that rundown of the hSlo channel observed on patch excision could be explained by cysteine oxidation, in part, because the reducing agent dithiothreitol (DTT) reversed the effect of patch excision to decrease the channel activity. This redox regulation may contribute to the observed variability in the voltage dependence of heterologously expressed hSlo (Stefani et al. 1997) and mSlo (Horrigan et al. 1999).

Methionine in proteins is readily oxidized to methionine sulfoxide (met(O)) (Vogt 1995), which is reduced back to methionine by the enzyme peptide methionine sulfoxide reductase (MSRA) using thioredoxin in vivo or DTT in vitro (Rahman et al. 1992; Moskovitz et al. 1996; Kuschel et al. 1999). Oxidation of methionine to met(O) by the addition of an oxygen atom drastically changes its side chain properties (Black and Mould 1991). Increasing evidence suggests that reversible oxidation of methionine involving MSRA may function as a general antioxidant mechanism (Levine et al. 1996; Moskovitz et al. 1998) and also as an important physiological regulator of many proteins (Ciorba et al. 1997, Ciorba et al. 1999; Berlett et al. 1998; Gao et al. 1998; Kuschel et al. 1999; Hoshi and Heinemann 2001). We showed previously that oxidation of methionine in voltage-dependent Shaker potassium channels regulates N-type inactivation (Ciorba et al. 1997) and P/C-type inactivation (Chen et al. 2000). Slowing of N-type inactivation induced by oxidation of a specific methionine residue in the NH₂-terminal ball domain was reversed by overexpression of MSRA (Ciorba et al. 1997; Kuschel et al. 1999). Oxidation of the NH₂-terminal methionine residue was promoted by NO donors and overexpression of nitric oxide synthase, suggesting that methionine oxidation could play a physiologically important role in the regulation of cellular excitability (Ciorba et al. 1999).

In the present study, we examined whether methionine oxidation regulates the function of hSlo channels heterologously expressed in mammalian cells. We show here that oxidation induced by chloramine-T (Ch-T) of methionine shifts the steady-state macroscopic conductance to a more negative direction by accelerating specific voltage-dependent opening transitions and also by slowing the rate limiting closing transition. Our results also suggest that the hSlo channel is regulated in an opposing manner by methionine oxidation and cysteine oxidation. This opposing regulation of the hSlo channel by cysteine and methionine oxidation contributes to the rundown and run-up of the channel induced by patch excision, and could play important roles in mediating the physiological and pathophysiological effects of ROS/RNS on cellular excitability.

The human Slo (U11058; Wallner et al. 1995) channel stably expressed in HEK cells (HF1; Meera et al. 1997) was obtained from the laboratory of Dr. R.W. Aldrich (Stanford University, Stanford, CA). This channel contains a myc tag at the NH₂ terminus, however, its electrophysiological properties are reported to be indistinguishable from those of the wild type hSlo channel (Meera et al. 1997). Similar electrophysiological effects of Ch-T were also obtained from HEK and COS cells transiently expressing the hSlo channel.

A mutant of the hSlo channel (huR2, U11058; Wallner et al. 1995) in which every cysteine was replaced with alanine is referred to here as cys-less hSlo and contains the following mutations: C14A, C53A, C54A, C56A, C141A, C277A, C348A, C422A, C430A, C485A, C498A, C554A, C577A, C612A, C615A, C628A, C630A, C695A, C722A, C797A, C800A, C820A, C911A, C975A, C995A, C1001A, C1011A, C1028A, and C1051A. The human clone huR2(+) (Wallner et al. 1995) was subcloned into pCI-neo vector (Promega). The cys-less huR2 DNA was generated by the “overlap extension” technique (Ho et al. 1989; Hirschberg et al. 1995). PCR amplification was carried out using proofreading Pfu DNA polymerase. Using the cys-less and wild-type channels, the following chimeric channels were constructed using the overlap extension method (Ho et al. 1989; Hirschberg et al. 1995): chimera No. 0 (C14A, C53A, C54A, C56A, C141A, C277A, C348A, C695A, C722A, C797A, C800A, C820A, C911A, C975A, C995A, C1001A, C1011A, C1028A, C1051A) and chimera No. 1 (C14A, C53A, C54A, C56A, C141A, C277A, C348A, C422A, C430A, C797A, C800A, C820A, C911A, C975A, C995A, C1001A, C1011A, C1028A, and C1051A). The integrity of the constructs was verified by nucleotide sequencing (automated DNA sequencer, ABI 377). The channels were transiently expressed in COS or HEK cells using GenePORTER™ 2 (Gene Therapy Systems). We did not observe any noticeable difference in the channel properties whether they are expressed in COS or HEK cells.

The Slo channel currents were recorded in the excised inside-out configuration using an AxoPatch 200A amplifier modified to expand the command voltage range or by an AxoPatch 200B (Axon). The output of the amplifier with the built-in filter set at 10 kHz was digitized using an ITC-16 AD/DA interface attached to an Apple Power Macintosh computer. The data acquisition was controlled by Pulse (HEKA). Linear leak and capacitative currents were subtracted using the P/n protocol as implemented in Pulse.

Patch pipets (Warner Instrument Corp.) were coated with dental wax and had a typical initial resistance of 0.8∼1 MΩ in the macroscopic current experiments. For the single-channel experiments, the pipet size was adjusted to obtain a small number of channels. In some experiments with a large number of channels (>10 nA at 180 mV), the series resistance was partially compensated (∼40%); however, in other experiments, the compensation was not routinely employed. The electrophysiological parameter values estimated did not show any systematic correlation with the current amplitude (see legends of Fig. 3 and Fig. 5) and suggest that the series resistance error had negligible effects on the results obtained. The experiments were performed at room temperature (20–23°C).

Both the external and internal solutions contained (in mM): 140 KCl, 2 MgCl₂, 11 EGTA, and 10 HEPES, pH 7.2, adjusted with N-methyl-d-glucamine. The free Ca²⁺ concentration of this solution was estimated to be <0.4 nM assuming that the residual contaminating Ca²⁺ concentration was 20 μM (Patcher's Power Tools v1.0, F. Mendez; http://www.wavemetrics.com/TechZone/User_ThirdParty/ppt.html) and the ratiometric Fura-2 measurement showed that this solution contained <0.2 nM free Ca²⁺. In some experiments, MgCl₂ was omitted to better study the Ca²⁺ sensitivity of the channel. Other solutions used are noted in the legends.

Chloramine-T (sodium salt), 5,5′-dithio-bis(2-nitrobenzoic acid) (DTNB) and p-chloromercuribenzoic acid (PCMB) were obtained from Sigma-Aldrich. Methanethiosulfonate ethylammonium (MTSEA) was purchased from Toronto Research Chemicals. The experimental solutions containing these reagents were prepared and the pH was adjusted immediately before use. Two types of experimental chambers were used in this study: one with an effective volume of ∼150 μl and the other with an effective volume of ∼500 μl. In those experiments, using the 150-μl chamber, typically Ch-T (four to five times the bath volume) was applied manually using a pipet and then washed out with 1 ml of Ch-T–free solution. The large chamber was continually perfused at 20 μl/s. We did not observe any systematic difference in the efficacy of Ch-T between the two methods. Recombinant bovine MSRA was purified from Escherichia coli as described previously (Moskovitz et al. 1996).

Macroscopic and single-channel current data were analyzed using PulseFit (HEKA), PatchMachine (http://www.hoshi.org), and IgorPro (WaveMetrics) running on Apple Power Macintosh computers. Statistical analysis was carried out using Data Desk (Data Description).

Ch-T is an oxidizing agent that preferentially oxidizes methionine to methionine sulfoxide (met(O)) and it also oxidizes cysteine (Shechter et al. 1975). Selective oxidation of methionine to met(O) by Ch-T (0.5–10 mM) has been demonstrated in different proteins (Shechter et al. 1975; Wang et al. 1985; Beck-Speier et al. 1988; Miles and Smith 1993; Nedkov et al. 1996; Schlief et al. 1996; Fu et al. 1998; Stief et al. 2000). Ch-T has been reported to modulate inactivation of several ion channels, including voltage-gated Na⁺ (Wang et al. 1985; Quinonez et al. 1999) and voltage-gated K⁺ channels (Schlief et al. 1996; Stephens et al. 1996; Ciorba et al. 1997), and at least some of these alterations have been attributed to oxidation of methionine residues (Schlief et al. 1996; Ciorba et al. 1997; Quinonez et al. 1999). To examine functional roles of methionine oxidation in the BK_Ca channel function, we investigated the effects of Ch-T on hSlo channels stably expressed in HEK cells (Meera et al. 1997). Representative macroscopic ionic currents obtained using the inside-out patch-clamp configuration in the virtual absence of [Ca²⁺] (<0.2 nM) are shown in Fig. 1. With this low internal [Ca²⁺], the Slo channel essentially acts as a voltage-dependent channel, thus, simplifying the data analysis (Meera et al. 1996; Cui et al. 1997; Horrigan and Aldrich 1999; Horrigan et al. 1999).

Application of 2 mM Ch-T to the cytoplasmic side in the inside-out configuration markedly increased the currents elicited by depolarization to 130 mV (Fig. 1 A). The peak hSlo current amplitude recorded at 130 mV as a function of time in one representative experiment is plotted in Fig. 1 B. In our experimental condition (<0.2 nM Ca²⁺), the current amplitudes were typically stable for 2–4 min after the patch excision. However, in longer recordings, the Slo macroscopic current amplitudes decreased in some patches (rundown; see Fig. 11 and Fig. 12 and also Dichiara and Reinhart 1997), whereas in other patches they actually increased (run-up; see Fig. 11 and Fig. 12). In nearly all of the patches examined, application of Ch-T to the cytoplasmic side gradually increased the peak current amplitudes recorded at 130 mV, usually by 200–300% (see Fig. 3 for statistics).

The effect of Ch-T to increase the hSlo current was not dependent on the expression level (see Fig. 3 legend for statistics). Similar effects were also observed when the hSlo α subunit alone was heterologously expressed in COS and HEK cells using the transient DNA transfection protocol. Although we cannot totally exclude the possibility that the effect of Ch-T was, in part, mediated by endogenous β subunits present in the cells tested, it is likely that Ch-T altered the Slo α subunit.

The enhanced hSlo current caused by Ch-T persisted even after the bath was washed extensively, up to 40 min. This observation is consistent with the idea that the channels were oxidatively modified by Ch-T. Furthermore, subsequent application of the membrane permeable reducing agent DTT (2 mM) to the patch did not reverse the current-enhancing effect of Ch-T (Fig. 1). Application of DTT alone readily reverses the effects of cysteine oxidation in many proteins, including voltage-gated K⁺ channels (Ruppersberg et al. 1991; Rettig et al. 1994; Heinemann et al. 1995). Therefore, this observation that application of DTT alone failed to reverse the effect of Ch-T to increase the Slo current amplitude suggests that reversible oxidation of cysteine may not mediate this phenomenon.

We found that the time course of the hSlo current increase induced by Ch-T was dependent on the Ch-T concentration used. With 2 mM Ch-T, the maximum current enhancement effect was usually observed within 1–2 min of application. With lower concentrations, the time course of the current enhancement was progressively slower and, with higher concentrations, the modification time course was faster. Because a prolonged incubation of the patch with Ch-T inevitably destroyed the seal, we typically used a brief application period (1–3 min) using 1–5 mM Ch-T to increase the Slo current amplitude and Ch-T was washed out. In the presence of higher concentrations of Ch-T (≥5 mM), a transient decrease in the Slo peak current amplitude was sometimes observed. This transient reduction in the current amplitude is consistent with a fast block of the channel by Na⁺ (Yellen 1984) as Ch-T was applied in the form of Na⁺ salt. We did find that 5 mM Na⁺ decreased the peak hSlo current amplitude by 20–30% (data not shown). However, we cannot exclude other possibilities such as Ch-T itself acting as an open channel blocker of the channel or Ch-T oxidizing other targets.

While Ch-T may oxidize cysteine in addition to methionine, the above observation that DTT fails to reverse the effect of Ch-T argues against the possibility that reversible cysteine oxidation is involved. To further verify this idea, we examined the effects of cysteine-specific reagents, DTNB (Eyzaguirre 1987; Shriver et al. 1998), MTSEA (Karlin and Akabas 1998), and PCMB (Eyzaguirre 1987; Shriver et al. 1998) on hSlo currents. In contrast to the current-enhancing effect of Ch-T, these sulfhydryl reagents all decreased hSlo currents in a qualitatively similar manner and application of DTT reversed the effect (Fig. 2 A), arguing that cysteine modification is unlikely to mediate the effect of Ch-T. Our results are consistent with those of Wang et al. 1997 who showed that MTSEA inhibited the opening of native BK_Ca channels in smooth muscle cells.

In addition, we also constructed a mutant hSlo channel in which all 29 cysteine residues were replaced with alanine (cys-less Slo). Although some hSlo-like macroscopic currents that were enhanced by Ch-T were detected in a small number of patches (n = 4), the overall expression efficiency in our practical experimental voltage range (up to 250–300 mV) was too poor to study effectively. Thus, we constructed a series of chimeric channels based on the wild-type and cys-less hSlo channels. Expression efficacies of the channels with mutations of the cysteine residues in and near the S7, S8, and S9 segments (C498, C554, C557, C612, C615, C628, and C630) were also too low to study in detail in a systematic manner (data not shown). Wood and Vogeli 1997 also reported that C612 may be important in the functional expression of the Slo channel in Xenopus oocytes using the two-electrode voltage-clamp method (Wood and Vogeli 1997).

Fig. 2 B shows two of the chimeric channels that could be readily examined electrophysiologically. These two channels in combination contain mutations of 21 out of 29 cysteine residues located in the core and the distal tail domains of the channel (C14, C53, C54, C56, C141, C277, C348, C422, C438, C485, C695, C722, C797, C800, C820, C911, C975, C1001, C1011, C1028, and C1051). Voltage dependence of the two chimeric channels was shifted to a more positive direction by at least 40 mV (compare Fig. 2 C with Fig. 3 D). Application of Ch-T markedly increased the currents through these channels (Fig. 2 C, top) in a manner similar to that observed with the wild-type channel (see Fig. 1). These mutant and cysteine reagent results together suggest that the effect of Ch-T to enhance the hSlo current, which is not reversed by DTT, is unlikely to be mediated by reversible oxidation of cysteine.

Representative hSlo currents recorded at different test voltages and the resulting peak current-voltage (I-V) curves obtained before and after Ch-T treatment are compared in Fig. 3. Ch-T increased the current amplitudes especially at slight and moderately depolarized voltages (80–160 mV), and the current-enhancing effect was less noticeable at very positive voltages (>180 mV; Fig. 3). The relative increase in the current amplitude as a function of the test voltage is presented in Fig. 3 C. The graph illustrates the potent effect of Ch-T to increase the hSlo channel currents at low and moderate voltages. At low voltages (90–120 mV), Ch-T often increased the current amplitudes by >500%. The voltage dependence shown in Fig. 3 C also confirms that Ch-T became less effective in enhancing the current amplitudes at more positive voltages where the open probability may be nearly saturated. This observation is consistent with the possibility that Ch-T increases the probability of the channel being open (P_o). Normalized macroscopic conductance-voltage (G-V) curves were obtained from the tail current amplitudes and compared in Fig. 3 D. We used a simple Boltzmann function as an operational measure, without any strict mechanistic implication, to describe the overall voltage dependence. The average G-V curve in the control condition was approximated by a simple Boltzmann function with an equivalent charge (Q_app) of 1.3e (±0.05e, n = 13) and the half-activation voltage (V_0.5) of 167 mV (±0.7 mV, n = 13, range 157–185 mV). After Ch-T treatment, V_0.5 dramatically shifted leftward by 29 mV to 138 mV (±2.9 mV, n = 13, ΔV_0.5 range 15–47 mV; P ≤ 0.0001 paired t test; Fig. 3 E), showing that at a given voltage, especially at low and moderately positive voltages, the channel is much more likely to be open. The average ΔV_0.5 illustrated in Fig. 3 may underestimate the maximum effect of Ch-T because the leftward V_0.5 shifts as large as 40–50 mV were sometimes observed with higher concentrations of Ch-T (5–10 mM) applied for longer durations (5–10 min). However, these harsh oxidizing treatments markedly increased the seal loss frequency.

The shape of G-V was also altered by Ch-T such that the curve appeared less steep. This effect is particularly more noticeable at very positive voltages (>160 mV) and at low voltages (<120 mV). Consistently, Ch-T treatment significantly decreased Q_app on the average by 0.2e (15%) from 1.3e to 1.1e (Fig. 3 E; P = 0.008, n = 13, paired t test).

Despite the obvious modification of the macroscopic G-V properties, the apparent reversal potential of the hSlo channel as estimated from the tail currents recorded at different voltages was unaltered by Ch-T (Fig. 4). This lack of change in the reversal potential, despite the readily observed shift in G-V (see above), argues against the possibility that Ch-T induced a large voltage shift in the recording system.

Cox et al. 1997a showed that an allosteric model with 10 states adequately explains the overall voltage-dependent mSlo gating modulated by Ca²⁺. According to this model, the results of Ch-T to enhance the channel activity at a given voltage and to induce a leftward shift in the macroscopic G-V curve could be explained by changes in the voltage-dependent open-closed equilibrium constant (L(V)) and/or the Ca²⁺ dissociation constants for the channel. However, the latter possibility is unlikely because the experiments described so far were conducted at a minimal [Ca²⁺] (<0.2 nM), and it may be assumed that the channels had essentially no Ca²⁺ ions bound (Horrigan et al. 1999). To further confirm that the action of Ch-T did not depend on the availability of divalent ions, we recorded hSlo currents in the presence of 10 mM EDTA without any added Ca²⁺ or Mg²⁺. We did not observe any noticeable difference in the voltage dependence of the channel activation when the internal solution contained EDTA. Even at these low divalent ion concentrations, Ch-T shifted the G-V curve to a more negative direction and increased the currents recorded at low voltages (data not shown), confirming that the action of Ch-T does not require Ca²⁺ or Mg²⁺.

Thus, the voltage-dependent open-closed equilibrium constant is the likely target of Ch-T. An allosteric model has been developed to describe the voltage-dependent gating of mSlo channel in the virtual absence of Ca²⁺ (Horrigan et al. 1999), and this model can be used as a framework to understand how Ch-T affects the voltage-dependent transitions of hSlo channel. To identify the gating transitions modified by Ch-T, we measured the activation and deactivation kinetics at the extreme voltages, which should reflect the rate limiting opening and closing transitions, respectively (Horrigan et al. 1999).

Application of Ch-T to the cytoplasmic side markedly slowed the kinetics of the hSlo tail currents. Representative scaled hSlo tail currents recorded at −40 and +40 mV before and after Ch-T treatment are shown in Fig. 5 A. We found that the tail current time course could be approximated by a single exponential at most of the voltages examined (but see Horrigan et al. 1999). The time constant values estimated at different voltages before and after application of Ch-T are plotted in Fig. 5 B. Ch-T increased the time constant of the tail current at every voltage examined, even at the most negative voltage examined (−150 mV). Horrigan et al. 1999 showed that the voltage dependence of the Slo deactivation time course has two distinct phases, one component with an equivalent charge of 0.52e and the other component moving 0.14e equivalent charges. Based on their results, we also fitted the voltage dependence of the hSlo tail deactivation kinetics, τ(V), by a sum of two exponential components (Fig. 5B) using the equation τ(V) = τ_A(0) · exp(z_AFV/RT) + τ_B(0) · exp(z_BFV/RT), where τ_A(0) and τ_B(0) are the time constant values of the two components at 0 mV, z_A and z_B are their respective equivalent charges, F is Faraday's constant, R is the gas constant, and T is the temperature. We found that oxidation by Ch-T specifically increased τ_A(0) and τ_B(0) (P = 0.0181, P = 0.0177, respectively, n = 10, paired t test) without significantly affecting their voltage dependence, z_A and z_B (P = 0.262, P = 0.8645, respectively, n = 10, paired t test). The mean values of τ_A(0) before and after application of Ch-T were 0.25 and 0.44 ms, respectively; and the mean values of τ_B(0) before and after were 0.14 and 0.61 ms, respectively. If it is assumed that the time constant values estimated at −150 mV reflect the single rate limiting closing transitions (Horrigan et al. 1999), the closing rate constant values before and after Ch-T are calculated to be 8.2 × 10³ ± 4.8 × 10² s⁻¹ and 5.5 × 10³ ± 4.1 × 10² s⁻¹, respectively.

In contrast to the marked effect of oxidation by Ch-T to slow the deactivation time course, Ch-T only slightly affected the activation time course of the Slo channel at very positive voltages where the open probability is relatively high. The hSlo currents recorded in response to pulses to 170 and 240 mV in the presence of <0.2 nM [Ca²⁺], before and after treatment with Ch-T, are shown in Fig. 6 A. At 170 mV, where the relative conductance is ∼0.6 (Fig. 3), the activation time course was faster after Ch-T treatment (Fig. 6 A, thick, dark sweep). However, at 240 mV, where the relative conductance is near saturation, the activation kinetics was not altered by Ch-T. Except for the initial few hundred microseconds after the depolarization onset, the currents were well described by a single exponential, which is consistent with the earlier observations (Cui et al. 1997; Horrigan et al. 1999). The time constant values estimated before and after Ch-T treatment are plotted as a function of voltage in Fig. 6 B. The results show that the difference in the activation time course observed at less positive voltages disappeared with greater depolarization. At 240 mV, where the open probability is nearly saturated, there was no statistical difference in the activation time constant before and after Ch-T treatment (P = 0.5645, n = 12, paired t test). Since the forward opening transitions dominate at these very positive voltages, oxidation by Ch-T should have no noticeable effect on the limiting opening transition. Based on the values of the time constant measured, the rate constant value (Horrigan et al. 1999) at 240 mV is estimated to be ∼550 s⁻¹ (control, 494 ± 75 s⁻¹; after Ch-T, 606 ± 72 s⁻¹). In a separate set of experiments, we also verified that Ch-T failed to accelerate the activation time course between 300 and 360 mV (n = 4).

Single-channel studies show that BK_Ca displays bursts of openings with short flicker closures involving multiple dwell time components (Barrett et al. 1982; Methfessel and Boheim 1982; Moczydlowski and Latorre 1983; Singer and Walsh 1987; McManus and Magleby 1988, McManus and Magleby 1991; Song and Magleby 1994; Rothberg and Magleby 1998, Rothberg and Magleby 1999). We investigated the effects of Ch-T on hSlo at the single-channel level. Representative single-channel openings recorded in the virtual absence of Ca²⁺ before and after Ch-T application are shown in Fig. 7 using two different time scales. Consistent with the macroscopic results, oxidation by Ch-T enhanced the overall mean current amplitude without obviously affecting the unitary current amplitude (Fig. 7 A).

Open duration histograms obtained from a representative patch are shown in Fig. 7 B. Ch-T increased the mean duration from 0.6 to 1 ms in this experiment, and the open duration distributions were statistically different (P < 0.01, Kolmogorov-Smirnov test). The results obtained from multiple experiments show that Ch-T increased the mean open duration typically by 50–100% in a statistically significant manner (P = 0.0006, n = 9, paired t test; Fig. 7 D). Similarly, the burst distributions shown were also significantly different (P < 0.01, Kolmogorov-Smirnov test). Ch-T increased the mean burst duration by 100–200% (P = 0.006, n = 9, paired t test; Fig. 7C and Fig. D). Thus, the longer open and burst durations at a given voltage contribute to the effect of Ch-T to increase the open probability and lead to greater Slo macroscopic currents.

Single-channel current-voltage (i-V) curves before and after Ch-T treatment were estimated using the ramp voltage protocol (Fig. 7 E). Only the open segments of the ramp current responses were averaged to construct composite open channel i-V curves. The results show that oxidation by Ch-T has no marked effect on the single-channel i-V, supporting the idea that the current-enhancing effect of Ch-T was caused by alterations in the hSlo channel gating to increase the open probability. This result is consistent with the observation presented earlier that the macroscopic instantaneous tail currents remained unaltered by Ch-T (Fig. 4).

The results presented above show that Ch-T shifts the voltage dependence of the macroscopic conductance and slows the deactivation kinetics without affecting the limiting activation kinetics in the virtual absence of Ca²⁺ (<0.2 nM). The macroscopic G-V of the Slo channel shifts to more negative voltages with increasing concentrations of Ca²⁺ (Jan and Jan 1997). We recorded the hSlo currents with different internal [Ca²⁺] to see how much [Ca²⁺] was required to induce the same amount of shift in V_0.5 as that caused by Ch-T in the virtual absence of Ca²⁺. We found that ∼300 nM [Ca²⁺] and Ch-T treatment induced about the same voltage shift in V_0.5 (Fig. 8, closed symbols). In addition, the time courses of the macroscopic currents recorded in 300 nM [Ca²⁺] before Ch-T, and in 0.2 nM [Ca²⁺] after Ch-T were similar (Fig. 8 B). These results suggest that oxidation by Ch-T may mimic almost a 1,000-fold increase in [Ca²⁺] from 0.2 nM (11 mM EGTA, no added Ca²⁺) to 300 nM to shift V_0.5 when α subunits are expressed alone.

Although the presence of Ca²⁺ is not required for the Ch-T action, we found that the shift in V_0.5 of G-V produced by Ch-T depended on Ca²⁺. We recorded the hSlo currents in the presence of 0.2 nM (Fig. 9, low Ca²⁺; 11 mM EGTA, no added Ca²⁺) and 120 μM Ca²⁺ (Fig. 9, high Ca²⁺), and examined the effects of Ch-T (Fig. 9). Representative Slo currents elicited at −60 and −20 mV in high Ca²⁺ and those at 120 and 180 mV in low Ca²⁺ are shown in Fig. 9 A. These voltages were selected because the relative conductance values were similar. At both −60 mV in high Ca²⁺ and 120 mV in low Ca²⁺, the relative macroscopic conductance is ∼0.1; and at both−20 mV in high Ca²⁺ and at 180 mV in low Ca²⁺, the relative conductance is roughly 0.6 (Fig. 9 B). At the voltages where the relative macroscopic conductance is 0.1 (−60 mV in high Ca²⁺ and 120 mV in low Ca²⁺), oxidation by Ch-T increased the current in high Ca²⁺ by ∼100%, whereas in low Ca²⁺, the current increased by >400%. At the voltages where the relative conductance was 0.6 (−20 mV in high Ca²⁺ and 180 mV in low Ca²⁺), Ch-T had no obvious effect on the current amplitude in high Ca²⁺, whereas in low Ca²⁺, it increased the current by 30%. As presented earlier, Ch-T markedly shifted the G-V curve leftward in low Ca²⁺ (0.2 nM; Fig. 9 B). The mean V_0.5 values before and after Ch-T treatment in low Ca²⁺ in this set of experiments were 164 and 128 mV (ΔV_0.5 = 36 mV), and the difference was statistically significant (P = 0.0008, n = 5, paired t test). In the presence of 120 μM Ca²⁺, Ch-T did significantly increase the hSlo currents at very low voltages (−80 and −60 mV). However, at more positive voltages, Ch-T was much less effective. The V_0.5 values estimated before and after Ch-T treatment in the presence of 120 μM Ca²⁺ were statically identical (−30 mV, P = 0.95, n = 5, paired t test). These results suggest that gating of the hSlo channel with high [Ca²⁺] in determining V_0.5 is not as oxidation-sensitive as that with low [Ca²⁺]. Whether the hSlo channel was treated with Ch-T in the presence of low or high [Ca²⁺] did not alter the subsequent electrophysiological properties measured after washing Ch-T out of the recoding chamber.

In the presence of high Ca²⁺, Ch-T did reduce Q_app by 0.42e or by 30% (P = 0.0037, n = 5, paired t test). This change in Q_app to decrease the steepness of G-V after Ch-T application manifests as greater conductance at low voltages (−80 to −40 mV). For example, Ch-T typically increased the current amplitude by 100% at −60 mV (Fig. 9 A, left) without increasing the currents at more positive voltage where the open probability is higher. This small reduction in Q_app is similar to that observed in the low Ca²⁺ condition (Fig. 3 E). Ch-T still slowed the deactivation time course in high Ca²⁺ (Fig. 9 A), even though V_0.5 was not significantly altered. These observations are consistent with the idea that Ch-T has multiple functional targets.

Cox et al. 1997a investigated the interaction of voltage and Ca²⁺ in controlling the mslo channel gating and showed that plotting the product of Q_app and V_0.5 as a function of [Ca²⁺] is a good experimental parameter to examine whether changes in the Ca²⁺ binding affinity or other conformational changes account for the apparent overall changes in the Ca²⁺ sensitivity of the Slo channel. The dependence of the product Q_app· V_0.5 on [Ca²⁺] before and after Ch-T treatment is shown in Fig. 9 C. Oxidation did not markedly affect the dependence of Q_app· V_0.5 on [Ca²⁺]. According to the formulation of Cox et al. 1997a, the results shown can be interpreted to indicate that Ch-T does not noticeably increase the affinity of Ca²⁺ binding itself, but alters other conformational changes to increase the open probability.

The observations presented thus far that Ch-T enhanced the hSlo current amplitude, that the effect was not reversed by the reducing agent DTT, and that Ch-T enhanced the mutant hSlo activity implicate the oxidation of methionine residues. The enzyme peptide methionine sulfoxide reductase (MSRA) specifically reduces met(O) back to methionine using thioredoxin in vivo or DTT in vitro (Moskovitz et al. 1996). MSRA in the presence of DTT has been shown to convert met(O) to methionine in the ShC/B ball peptide and restore fast N-type inactivation (Ciorba et al. 1997). We tested whether MSRA could reverse the effect of Ch-T to enhance the hSlo activity by applying recombinant bovine MSRA (Moskovitz et al. 1996) to the cytoplasmic side of the patch (Fig. 10). First, Ch-T was applied to increase the Slo channel activity and the subsequent DTT (2 mM) application did not reverse the effect. However, application of MSRA (15 μg/ml) in the presence of DTT (2-5 mM) gradually decreased the current amplitude, partially reversing the effect of Ch-T. This concentration of DTT is less than optimal for the enzymatic activity in vitro (Moskovitz et al. 1996), however, higher concentrations of DTT almost immediately destroyed the seal in our experimental condition. In addition to the effect on the peak amplitude, MSRA application also accelerated the tail current time course. Similar results were obtained in five other patches. Application of the enzyme alone without any added DTT produced no consistent effect (data now shown; n = 3). It is not clear why we were unable to achieve complete reversal. It is possible that the residues oxidized by Ch-T are not readily accessible by the enzyme. For example, some of the target methionine residues may face the extracellular side. Another possibility, for only the partial reversal, may be that methionine was oxidized by Ch-T to methionine sulfone by the addition of two oxygen atoms. Ch-T oxidizes the methionine residue in the ShC/B ball peptide (Ciorba et al. 1997) to methionine sulfone (Heinemann, S.H., and T. Hoshi, unpublished results). MSRA is not effective in reducing methionine sulfone (Eijiri et al. 1979).

The action of MSRA to at least partially reverse the effect of Ch-T further indicates that oxidation of methionine enhances the Slo channel activity. Furthermore, because the enzyme is not likely to cross the membrane from the intracellular side to the extracellular side, the results suggest that at least some of the oxidation and the enzyme target residues must face the cytoplasmic side.

Gating properties of the Slo channel are known to change over time, especially in the excised-patch configuration (Silberberg et al. 1996; Dichiara and Reinhart 1997). Electrophysiological properties of many other ion channels also change after patch excision, and a variety of biophysical and molecular mechanisms have been proposed to account for these changes (Horn and Korn 1992; Becq 1996). Dichiara and Reinhart 1997 showed that the hSlo current amplitude decreases with time on patch excision, and that this rundown is prevented by the reducing agent DTT and mimicked by oxidation induced by H₂O₂. They concluded that cysteine oxidation might be involved. We also observed that in some patches, the hSlo current amplitude gradually decreased after patch excision (Fig. 11 A). Consistent with the results of Dichiara and Reinhard (1997), we found that H₂O₂ frequently promoted and DTT reversed the channel rundown (data not shown). We also found that DTNB, which preferentially modifies cysteine residues, was effective in decreasing the hSlo current amplitude, and this effect was readily reversed by application of the reducing agent DTT (Fig. 11 B; also see Fig. 2). In contrast to the effect of Ch-T, DTNB did not affect the deactivation time course in a consistent way, suggesting that distinct biophysical mechanisms underlie the effects of these oxidizing agents (data not shown).

In other patches, hSlo current amplitudes actually increased with time after patch excision (run-up; Fig. 11 C). Based on the foregoing observations, we hypothesized that the increase in the hSlo current amplitude after patch excision might involve methionine oxidation. If so, application of MSRA, which catalyzes the reduction of met(O) to methionine, should reverse this effect. In contrast to the effect of DTT on the channel rundown, application of DTT (2 mM) did not alter the increased Slo current amplitude after patch excision (Fig. 11 C). The small increase in the current amplitude seen with the application of DTT may represent a reversal of concurrently occurring rundown mediated by reversible cysteine oxidation (also see Fig. 1 B and Fig. 10 B). The failure of DTT to decrease the current argues against the involvement of cysteine oxidation, which should be easily reversed by DTT (see above). Application of purified MSRA (15 μg/ml) together with DTT (2 mM) to the cytoplasmic side of the patch decreased the run-up Slo current to the initial current level observed immediately after patch excision (Fig. 11 D). Similar observations were obtained in five other macroscopic patches. Thus, spontaneous run-up of hSlo induced by patch excision is likely caused by oxidation of methionine residues accessible from the cytoplasmic side. We found that MSRA was noticeably more effective in decreasing the currents which underwent spontaneous run-up after patch excision than the currents enhanced by Ch-T. This is consistent with the possibility that Ch-T may oxidize methionine not only to met(O) but also to methionine sulfone, which is not reduced by MSRA (Eijiri et al. 1979).

The results shown above can be used to explain fluctuations of the macroscopic Slo current amplitude observed in some patches in the excised configuration. Fig. 12 shows the Slo current amplitudes as a function of time after patch excision in three different patches. One patch showed prominent rundown, another patch showed run-up, and the third patch showed slow oscillations in the current amplitude. Out of >140 patches examined in the virtual absence of Ca²⁺ (0.2 nM), the patches that showed rundown and run-up were relatively common, ∼50 and 20%, respectively. The patches that exhibited the current fluctuations were much less common (<5%). In the remaining 25%, the current amplitudes were stable. These fluctuations could be considered as a series of rundown and run-up phenomena, which, in turn, could be explained by cysteine and methionine oxidation, respectively.

The results presented above suggest that, in the hSlo channel-containing patches, oxidation of methionine residues accessible from the cytoplasmic side by Ch-T and at least partially by MSRA enhances the channel activity. The hSlo channel contains many methionine residues, especially in the cytoplasmic tail domain, which are thought to contribute to the Ca²⁺ sensitivity of the channel (Wei et al. 1994; Schreiber and Salkoff 1997; Schreiber et al. 1999). The S5/P/S6 segments of the channel also contain multiple methionine residues. If the methionine residues located within the channel pore are the Ch-T targets, it might be possible to protect the channel from oxidative modification by using the channel blockers that hinder Ch-T's access to the target sites. Therefore, we examined the effect of intracellular TEA on the ability of Ch-T to enhance the hSlo channel activity. In Shaker voltage-dependent K⁺ channels (ShB), M440, T441, and T469 located in or near the pore cavity act as major determinants of internal TEA binding (Yellen et al. 1991; Choi et al. 1993).

The results from a typical experiment using TEA and Ch-T are shown in Fig. 13. The hSlo channel is less sensitive to internal TEA than the ShB channel and higher concentrations were necessary to block the channel activity. Application of 80 mM TEA to the cytoplasmic side decreased the macroscopic hSlo current at 140 mV to ∼10% of the control level. Then, concurrently with TEA, Ch-T was applied to the patch and the channels were incubated in the presence of TEA and Ch-T together for 1.5 min, which is sufficiently long for Ch-T to noticeably increase the hSlo current when applied alone (see Fig. 1). In the presence of both TEA and Ch-T, the current amplitude remained at ∼10% of the control level, which is consistent with the interpretation that Ch-T did not markedly alter the efficacy of TEA to block the channels. After the concurrent application of TEA and Ch-T together, the bath was washed free of these agents. The hSlo current amplitude quickly recovered but only to 10–15% above the control level before the TEA application, and thereafter the current amplitude remained stable. If Ch-T had modified the channels while they were blocked by TEA, the current amplitude after washing out TEA and Ch-T should have been markedly greater than the control amplitude. The deactivation time course was essentially unchanged during the concurrent application of TEA and Ch-T, and subsequent application of Ch-T alone without TEA slowed the time course. These observations suggest that Ch-T failed to modify the channels while they were blocked by TEA, and are also consistent with the possibility that the channel protein itself is directly oxidized by Ch-T. Subsequent application of Ch-T without TEA progressively increased the hSlo current amplitude. Similar results were obtained in six other experiments using TEA. It is not clear why the second Ch-T application without TEA induces only a 40–50% increase in the current amplitude. When applied alone without prior application of TEA, Ch-T often induces a 100–200% increase (Fig. 2). Multiple oxidation targets and their allosteric interactions may be involved. In addition to TEA, we also used another blocker of the hSlo channel that works at much lower concentrations (MPTP [1-methyl-4-phenyl-1, 2, 3, 6-tetrahydropyridine], 225 μM; Tang, X.D., and T. Hoshi, manuscript in preparation) and obtained the same results.

Voltage-dependent activation involves conformational changes of ion channel proteins and accessibility/reactivity of the amino acid residues may change during the activation process (Yellen 1998). Thus, we examined whether the effect of Ch-T to enhance the Slo channel activity was regulated by channel opening/closing by manipulating the holding voltage while the channels were exposed to Ch-T. Keeping the patch at −100 mV without any depolarization or at 130 mV without repolarization at 0.2 nM [Ca²⁺] did not interfere with the effect of Ch-T to increase the current amplitude.

Application of oxidizing agents produces multiple effects on the hSlo channel. The results presented in this study show that, in the virtual absence of Ca²⁺, Ch-T increased the Slo channel current by shifting its G-V to a more negative direction by ∼30–50 mV. This effect was partially reversed by MSRA and DTT applied together but not by DTT alone. Those reagents that specifically modify cysteine, in contrast, decreased the hSlo channel current.

Most naturally occurring amino acids in proteins can be oxidized, however, only the oxidation of cysteine and methionine is readily reversible. Oxidized cysteine is easily reduced back by the reducing agent DTT. For example, N-type inactivation in Kv1.4 and Kv1.4/Kvβ is slowed by oxidation in a cysteine-dependent manner and this effect is readily reversed by DTT (Ruppersberg et al. 1991; Rettig et al. 1994; Heinemann et al. 1995). The observation that the decreased Slo channel current induced by patch excision and H₂O₂ is restored by DTT implicates cysteine oxidation as the underlying mechanism. This conclusion is consistent with the earlier observation of DiChirara and Reinhart (1997). Similarly, DTNB, MTSEA, and PCMB, cysteine-specific reagents, decreased hSlo currents and their effects were reversed by DTT.

Upregulation of the hSlo channel activity induced by Ch-T and by patch excision in some experiments is likely to involve methionine oxidation. Our results show that DTT alone does not reverse the increased Slo channel current, but that MSRA applied concurrently with DTT partially reverses the increased current induced by Ch-T or spontaneous run-up, suggesting that oxidation of methionine to met(O) is involved. MSRA specifically reduces met(O) back to methionine using cellular thioredoxin or DTT in vitro (Rahman et al. 1992; Moskovitz et al. 1996). Several possibilities exist to account for the partial but not full recovery obtained with MSRA. Some of the methionine residues could have been indeed oxidized to methionine sulfone by the addition of two oxygen atoms by Ch-T. MSRA does not reduce methionine sulfone (Eijiri et al. 1979). Another possibility to account for the partial recovery by MSRA is that the oxidized methionine residues may not be easily accessible to the enzyme, as would be the case if they are located near or in the pore cavity. It is also possible that in our experimental condition, the enzyme activity was not sufficiently high (Moskovitz et al. 1996). The observation that all the cysteine-specific reagents we examined decreased the Slo current is consistent with the idea that oxidation of methionine, not cysteine, underlies the effect of Ch-T and run-up. Furthermore, the stimulatory effect of Ch-T is retained in the mutant channels that lack a majority of the cysteine residues (21/29), especially those in the core domain. The robust and consistent effect of Ch-T observed in cell-free excised patches suggests that methionine residues in the hSlo channel are directly oxidized to alter the channel gating. Similar effects of Ch-T were observed when hSlo was expressed in COS and HEK cells and it is not likely that endogenous β subunits markedly contributed to the phenomena reported here. The finding that internal TEA protects the channel from oxidative modification by Ch-T (Fig. 13) is consistent with the interpretation that the functionally important residues oxidized by Ch-T are located in the channel itself, most likely in the S5/P/S6 segments.

Oxidation of cysteine and methionine, therefore, has opposing effects on the overall Slo channel function. These mechanisms appear to contribute to the rundown, run-up, and current fluctuations of hSlo channels observed after patch excision. However, it should be noted other mechanisms could induce similar up- and downregulation of the hSlo channel activity, and it is presumptuous to assume that the up- and downregulation of the channel is solely mediated by methionine and cysteine oxidation. Although cysteine and methionine oxidation appear to have opposing overall effects on the hSlo channel current, it is not likely that these mechanisms alter the same gating transitions in the opposite directions. Methionine oxidation increases the open probability without markedly affecting the activation time course (this study), whereas cysteine oxidation noticeably slows down the activation time course (Dichiara and Reinhart 1997).

Several models have been developed to account for gating of BK_Ca channels (Horrigan et al. 1999; Nimigean and Magleby 1999; Talukder and Aldrich 2000). Here, we have opted to use the model developed by Horrigan et al. 1999(also see their Table I) to account for the observed results of methionine oxidation induced by Ch-T. For the sake of simplicity, the analysis presented here assumes that all the channels measured are homogeneous in terms of the redox state. It is possible that some of the variability observed before application of Ch-T could be explained by differential levels of basal oxidation, thus, violating the homogeneity assumption. This allosteric gating model (referred to here as the HCA model) postulates that the channel operates in an allosteric manner involving five closed (C₀–C₄) and five open states (O₀–O₄) as shown in Scheme I.

The transitions among the closed and those among the open states (the horizontal transitions in (Horrigan et al. 1999)) are faster and more voltage-dependent than the rate limiting closed-open transitions (the vertical transitions in (Horrigan et al. 1999)). In response to depolarization, the channel may typically proceed from C₀, C₁, C₂, C₃, C₄, and then to O₄. Upon repolarization to a very negative voltage, the channel is likely to close following the O₄-O₃-O₂-O₁-O₀-C₀ pathway. The activation time course at very positive voltages primarily reflects the C₄-O₄ transition, whereas the deactivation time course at negative voltages is mainly determined by the O₀-C₀ transition.

We found that only minor adjustments in the average parameter values in the HCA model at 20°C (Horrigan et al. 1999) were necessary for adequate simulation of the hSlo gating behavior in our control condition (<0.2nM Ca²⁺; see Fig. 14 legend for the parameter values). The value of the closing rate constant γ₀ can be estimated from the deactivation kinetics at −150 mV and the value of the opening rate constant δ₄ from the activation kinetics at 240 mV. The voltage dependence of the rate constants are assumed to be the same as those given in Horrigan et al. 1999 because the voltage dependence of the steady-state macroscopic conductance, the activation kinetics, and deactivation kinetics measured in our study are similar to those obtained by Horrigan et al. 1999. The α(0) and β(0) values can be estimated by fitting Eq. 13 in Horrigan et al. 1999 to the measured G-V, assuming that the value of L(0) for hSlo is the same as that for mSlo.

The parameter values were determined in a similar manner and further adjusted to simulate the effects of Ch-T. The following criteria were used to evaluate the parameter adjustments. First, V_0.5 shifts to more negative voltages by 30 mV (Fig. 3D and Fig. E). Second, Q_app decreases slightly (Fig. 3D and Fig. E). Third, the activation time course at very positive voltages is not accelerated (Fig. 6). Fourth, the deactivation kinetics is slower (Fig. 5). With the values of δ₄ and γ₀ constrained as discussed above, increasing the allosteric factor D shifted V_0.5, leftward, however, unlike the effect of Ch-T, it also increased Q_app. A similar noticeable but small increase in Q_app was observed when the ratio δ₀(0)/γ₀(0) was varied to simulate the leftward shift in V_0.5. These considerations make any alterations in the allosteric factor D or the ratio δ₀(0)/γ₀(0) unlikely to underlie the observed effects of Ch-T. Because changes in δ₀, δ₁, δ₂, and δ₃ do not markedly alter the macroscopic G-V, activation or deactivation kinetics as verified by model simulations, we left these parameters unchanged. With these constraints, we found that acceleration of the C₀-C₁-C₂-C₃-C₄/O₀-O₁-O₂-O₃-O₄ opening transitions (the two horizontal transitions in Horrigan et al. 1999 by increasing the value of α at 0 mV, α(0), by a 2.3-fold (ΔΔG ≈ 2.1 J/mol) was necessary to simulate the shift in V_0.5 caused by Ch-T. This simulated shift was also accompanied with a small decrease in Q_app. A 60% decrease in the value of the O₀-C₀ transition or γ₀ was required to simulate slowing of the deactivation time course. The C₄-O₄ transition or δ₄ was left unchanged to preserve the activation kinetics at very positive voltages. The hSlo macroscopic currents and the G-V curves simulated using the parameter values estimated as described above are shown in Fig. 14. The results nicely meet the four criteria listed above to account for the effects of Ch-T. Although the single-channel properties were not used as the evaluation criteria, the changes in the model parameters well reproduced the increases in the mean open and burst durations observed (data not shown).

The overall voltage dependence of the hSlo channel is regulated by intracellular Ca²⁺ within the dynamic range of 100 nM to 1 mM (Vergara et al. 1998). Ca²⁺ binding is often considered to allosterically modulate the voltage-dependent activation (Cox et al. 1997a; Cui et al. 1997). In the virtual absence of Ca²⁺, methionine oxidation induced by Ch-T shifts V_0.5 by 30–50 mV, functionally mimicking the effect of 300–400 nM [Ca²⁺] (Fig. 8). However, this shift in V_0.5 by oxidation diminishes in the presence of 120 μM [Ca²⁺], which is near the high limit of the Ca²⁺ dynamic range of the channel. Although the overall effect of methionine oxidation on V_0.5 is dependent on [Ca²⁺], the analysis using the product V_0.5 · Q_app (Cox et al. 1997a) fails to support the idea that oxidation directly alters the Ca²⁺ binding affinity of the channel. Thus, one attractive but speculative idea is that methionine oxidation in the hSlo channel partly mimics the influence of the Ca²⁺ binding domain on the voltage-dependent activation mechanism. Even in high [Ca²⁺], Ch-T does increase the currents at moderately positive voltages by decreasing Q_app and slows the deactivation kinetics. This suggests that methionine oxidation may regulate functional roles of BK_Ca at a variety of [Ca²⁺] concentrations.

The Slo channel contains a large number of methionine residues and identification of the residues oxidized to bring about the functional effects reported here will require further investigation. However, our results do provide some clues as to where the target residues may be located in the Slo channel. Our simulations using the HCA model show that acceleration of the C₀-C₁-C₂-C₃-C₄/O₀-O₁-O₂-O₃-O₄ opening transitions may underlie the effect of methionine oxidation to shift the voltage dependence of the channel. Horrigan et al. 1999 suggest that these opening transitions may represent the activation of voltage sensors. Given this interpretation, it could be speculated that methionine residues in the S4/S4–S5 linker segments may be involved. However, the hSlo and mSlo channels do not contain any methionine residue in these segments and the target residues are likely located elsewhere. The result that the intracellular channel blockers protect the Slo channel from functional modification by Ch-T suggests that the target residues may lie near the channel pore. Because MSRA at least partly reverses the effect of Ch-T, the targets must be accessible from the cytoplasmic side and the residues located deep in the pore can be excluded from consideration. In Shaker channels, the PVP motif in the S6 segment may introduce a sharp bend in the S6 segment such that those residues located downstream of this motif may be well exposed to the cytoplasmic side (Durell and Guy 1992; del Camino et al. 2000). In hSlo and mSlo channels, the sequence PVP is replaced with YVP, and it is not clear whether a similar arrangement exists. If it does, the methionine residues located near or carboxyl to the YVP motif are the prime oxidation targets. In the simulation presented above, the values of two rate constants in the HCA model, α(0) and γ₀(0), were adjusted to account for the shift in V_0.5, reduced Q_app and slowing of the deactivation kinetics. It is not clear at present whether oxidation of a single methionine residue or multiple residues accounts for the required changes in the two rate constants. One scenario that needs to be considered is that for Ch-T to alter the channel function, oxidation of multiple residues in series is required. Ch-T may oxidize one residue, which increases the likelihood of the second residue becoming oxidized, perhaps by increasing the accessibility. The alterations in the multiple rate constants required to adequately simulate the results may reflect this possible sequential nature of the modification process. This mechanism may also explain why the current amplitude continued to increase even after Ch-T was washed out in some experiments (Fig. 1 B).

Taken together, the following tentative model of the action of methionine oxidation to facilitate the Slo channel opening may be proposed. It may be argued that the critical methionine residues in the pore/S6 segments, which are exposed to the cytoplasmic side but sheltered when an intracellular pore blocker is present, are involved in integrating the information from the voltage sensors and the Ca²⁺ binding domain, contributing to regulation of the activation gate. Oxidation of these methionine residues to met(O) by the addition of an extra oxygen atom to the sulfur atom stiffens the side chain and renders it markedly more polar. The hydrophilicity of met(O) has been estimated to be similar to that of lysine (Black and Mould 1991). This change in the side chain property may partially open the activation gate, likely to be comprised in part by the residues in the S6 segment (Liu et al. 1997; del Camino et al. 2000), and by twisting the S6 segment, as shown for the inner TM2 helices in the KcsA channel (Perozo et al. 1999). This partial opening or potential twisting of the S6 segment primes the activation gate for opening, and it is equivalent to 30–50 mV depolarization (ΔΔG ≈ 2.1 J/mol) or increasing [Ca²⁺] from 0.2 to 300 nM.

Ca²⁺-dependent K⁺ channels play multitudes of physiological roles, ranging from cochlear frequency tuning (Fettiplace and Fuchs 1999; Ramanathan et al. 1999) to vascular tone control (Brayden 1996; Rusch et al. 1996; Brenner et al. 2000; Pluger et al. 2000). Free radicals, such as hydroxyl radicals, capable of promoting oxidation of amino acids are also implicated in vascular regulation (Rubanyi 1988). NO, an important regulator of blood vessel function (Drexler 1999; Zanzinger 1999; Sanders et al. 2000), is a weak radical and may directly or indirectly promote oxidation (Stamler 1994). Ciorba et al. 1999 showed that NO could promote methionine oxidation in Shaker channels, and it would be interesting to see if NO could increase hSlo currents. Some studies have shown that application of oxidizing agents increased the BK_Ca channel activity (Thuringer and Findlay 1997; Barlow and White 1998; Hayabuchi et al. 1998; Gong et al. 2000), whereas others have shown inhibition (Dichiara and Reinhart 1997; Wang et al. 1997; Brzezinska et al. 2000). Especially in intact cells, it is difficult to predict whether application of oxidizing agents will increase or decrease the BK_Ca channel activity. Oxidation may have direct effects on the channel protein mediated by oxidation of its amino acid residues and also indirect effects mediated by oxidation of other molecules, such as those involved in intracellular Ca²⁺ homeostasis (Suzuki et al. 1997; Kourie 1998; Chakraborti et al. 1999). The presence of β subunits in some, but not all, Ca²⁺-dependent K⁺ channel complexes further contributes to the complexity (Chang et al. 1997; Tanaka et al. 1997; Wanner et al. 1999). For example, in our study, DTNB, like other cysteine-specific reagents examined, decreased the hSlo current, whereas native rat hippocampal BK_Ca currents were enhanced by DTNB and reversed by DTT (Gong et al. 2000). Even considering only the direct action on the channel protein itself, the effects of oxidation could be complex since there may be multiple oxidation targets, most likely cysteine and methionine residues. The results presented here show that, in the heterologously expressed Slo channel, cysteine oxidation generally inhibits, whereas methionine oxidation enhances the BK_Ca channel activity. It will be interesting to know the relative efficacies of cysteine and methionine oxidation in vivo.

Unlike some other regulatory molecules such as protein kinases, ROS/RNS, capable of oxidizing amino acid residues, do not have specific target amino acid consensus sequences for their action and the potential specificity of their action may be conferred by other mechanisms. Possible colocalization of the molecules that generate ROS/RNS with the target molecules and the availability of cofactors, such as Fe, may contribute to the potential specificity.

Clinically, the results presented here may be relevant to reperfusion injury. During reperfusion after an ischemic episode, a burst of free radicals is produced, promoting oxidation of amino acids (Babbs 1988; Rubanyi 1988; Gress 1994). Vascular BK_Ca channels may be oxidatively modified during reperfusion. Methionine oxidation of BK_Ca channels, which facilitate repolarization to restrict excess Ca²⁺ entry, may serve as a neuronal protective mechanism. Aberrant elevations in intracellular Ca²⁺ and ROS/RNS also have been documented in aged animals and some neurodegenerative diseases (Olanow and Arendash 1994). When the enhanced BK_Ca channel activity is no longer desired, MSRA could decrease the channel function by reducing met(O) to methionine. It is interesting to note that a decrease in the MSRA activity was reported in the Alzheimer's disease brain (Gabbita et al. 1999), which involves increased oxidative stress (Markesbery 1997). Therefore, methionine oxidation and MSRA could function as a reversible cellular regulatory mechanism.

In summary, we showed here that Ch-T via methionine oxidation increases the hSlo Ca²⁺-dependent K⁺ channel activity and that methionine oxidation and cysteine oxidation have opposing effects on the channel activity. We suggest that methionine oxidation may have a role in integration of the information from the voltage sensors and the Ca²⁺ binding domain of the channel. Considering that ROS/RNS production is closely linked to cellular metabolism, reversible regulation of the Slo channel by methionine oxidation and MSRA may represent an important functional coupling mechanism between cellular metabolism and electrical excitability.

We thank C. Schinstock for technical assistance, Dr. Yermolaieva for the fura-2 Ca²⁺ measurement, Drs. V. Avdonin and E. Shibata for discussion and reading of the manuscript, and H. Sumlin for encouragement.

This study was supported in part by the National Institutes of Health grants GM57654 and HL14388 (to T. Hoshi).